CN103439708A - Polarized InSAR interferogram estimation method based on generalized scattering vector - Google Patents

Abstract

The invention discloses a polarized InSAR interferogram estimation method based on a generalized scattering vector, and mainly solves the problem that an existing polarized InSAR interferogram generating algorithm is unstable in registration error. The polarized InSAR interferogram estimation method is implemented in the following steps of (1) inputting image data, (2) roughly registering images, (3) establishing the generalized scattering vector, (4) selecting image windows, (5) estimating a generalized coherence matrix, (6) estimating a generalized interference matrix, (7) establishing an eigen decomposition matrix, (8) decomposing matrix eigens, (9) arranging eigen values, (10) generating interferometric phases, (11) judging whether all the interferometric phases are obtained, and (12) obtaining an interferometric phase diagram. The polarized InSAR interferogram estimation method based on the generalized scattering vector has the advantage of being capable of obtaining a high-quality interferogram even on the condition that a polarized channel and/or a space channel has registration errors, can effectively reduce influence of the registration errors on the interferogram, and fully utilizes all polarization information of each pixel.

Description

Polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector

Technical field

The invention belongs to communication technical field, further relate to the polarization interference synthetic aperture radar based on the broad sense Scattering of Vector (Synthetic Aperture Radar Interferomety, InSAR) the interferogram method of estimation in the radar exploration technique field.The present invention carries out the generalized character decomposition to the pixel in radar image, then weighting obtains interferometric phase image to original interferogram.The present invention can be used in POLARIZATION CHANNEL and/or the non-complete registration situation of spatial channel, the sane estimation of polarization InSAR interferogram.

Background technology

Polarization InSAR is the study hotspot of modern radar, it had both kept the measuring characteristic of InSAR, introduced again polarimetric synthetic aperture radar (Synthetic Aperture Radar, SAR) measurement advantage, so can obtain polarization scattering characteristics and the interference information of atural object, and can improve by extra information the application performance of polarization SAR and InSAR simultaneously.

Interferometric phase image is that after the major-minor connection of antenna of radar is received respectively to the phase place of signal is done the interference processing, the width obtained means the image of two antenna receiving signal phase differential.It is in the processing of polarization InSAR that the polarization Coherence optimization decomposes, and finds a kind of polarized state combination, makes coefficient of coherence maximum process under this polarized state combination.The key factor that impact polarization Coherence optimization decomposes and interferometric phase image generates is Images Registration, and image registration accuracy has directly determined the quality of interferometric phase image result.The existing interferometric phase image method of estimation based on Pauli base Scattering of Vector, when the image accuracy registration, can obtain interferometric phase image preferably and estimate; When between image, having registration error, the interferometric phase image estimated performance descends, and phase noise increases; When having larger registration error, even can not obtain correct interferogram estimated result.

The registration process of existing polarization InSAR mainly be take piece image as reference picture, and remaining image all carries out registration with reference picture, and registration results is applied in each image.

Xiong Tao is at article " the gordian technique research of polarization interference synthetic aperture radar application " (Tsing-Hua University's PhD dissertation, 2009) introduce Similarity Parameter in the polarization interference data are carried out to registration, Tan Lulu and Yang Ruliang " utilize Cameron to decompose and carry out the right Study of Registration of polarization interference image " (systems engineering and electronic technology at article, 2009,31 (6): 1284-1287), also proposed to utilize Cameron to decompose to carry out the registration of polarization interference image.Said method is greatly improved with respect to traditional method for registering registration accuracy.Yet the weak point of said method is: when carrying out registration, suppose that between the POLARIZATION CHANNEL of image be accuracy registration, yet under normal conditions, due to the impact that is subject to the factors such as coarse antenna phase center position and the incomplete quadrature of POLARIZATION CHANNEL, through between calibrated POLARIZATION CHANNEL, still there is residual error.

The patented claim that Civil Aviation University of China proposes " the InSAR interferometric phase method of estimation based on related weighing associating the subspace projection " (date of application: on 01 09th, 2012, application number: 201210003415.5, publication number: disclose a kind of method based on related weighing associating subspace projection CN102565798A) and carried out sane estimation interferometric phase.At first the method utilizes existing correlation method to carry out thick registration to major-minor image, then constructs related weighing associating covariance matrix and weighting associating correlation matrix, and the cost function that finally utilizes the noise subspace projection to construct estimates interferometric phase.The method can obtain interferometric phase result accurately in the situation that registration accuracy is very poor.The weak point that the method exists is: for the data of just single polarization InSAR process, and, when processing multipolarization InSAR data, because there is residual error between POLARIZATION CHANNEL and/or spatial channel, the method can obtain undesirable interferometric phase image.

Summary of the invention

The polarization InSAR interferogram generation technique that the present invention is directed to above-mentioned prior art existence utilizes the defect of insufficient deficiency to polarization information, a kind of polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector has been proposed, by setting up the broad sense scattering model, and reduce POLARIZATION CHANNEL and/or spatial channel registration error to the polarization Coherence optimization decomposes and interferometric phase image generates impact based on this model, while having guaranteed to have residual error between POLARIZATION CHANNEL and/or spatial channel, still can obtain the quality InSAR interferometric phase image that polarizes preferably.

For achieving the above object, key step of the present invention is as follows:

(1) input image data:

Main image data 1a) the polarization interference synthetic aperture radar main antenna obtained is input to system;

Auxiliary view data 1b) the auxiliary antenna of polarization interference synthetic aperture radar obtained is input to system;

(2) the thick registration of image:

Utilize the thick method for registering of InSAR, master image and auxiliary image are carried out to thick registration process;

(3) build the broad sense Scattering of Vector:

3a) choose arbitrarily a pixel in master image, element in the Poly Pauli base Scattering of Vector of neighbor pixel around selected pixel and its is formed a line, construct the broad sense Scattering of Vector of selected pixel in master image;

3b) choose successively the pixel that not yet builds the broad sense Scattering of Vector in master image, by all pixels of choosing successively with the Poly Pauli base Scattering of Vector of neighbor pixel around it in element form a line, construct the broad sense Scattering of Vector of all pixels of choosing in master image;

Whether the broad sense Scattering of Vector that 3c) judges all pixels in master image has all built, if all built, performs step 3d), otherwise, execution step 3b);

3d) choose arbitrarily a pixel in auxiliary image, element in the Poly Pauli base Scattering of Vector of neighbor pixel around selected pixel and its is formed a line, construct the broad sense Scattering of Vector of selected pixel in auxiliary image;

3e) choose successively the pixel that not yet builds the broad sense Scattering of Vector in auxiliary image, by all pixels of choosing successively with the Poly Pauli base Scattering of Vector of neighbor pixel around it in element form a line, construct the broad sense Scattering of Vector of all pixels of choosing in auxiliary image;

Whether the broad sense Scattering of Vector that 3f) judges all pixels in auxiliary image has all built, if all built, and execution step (4), otherwise, execution step 3e);

(4) choose image window:

In master image, choose a pixel as center, take regular length as the piece radius, choose a foursquare window, in auxiliary image, choose with pixel corresponding in master image as center, obtain onesize square window;

(5) estimate the broad sense coherence matrix:

5a) in the master image window, after being multiplied each other, the broad sense Scattering of Vector of all pixels in this window and the conjugate transpose of self get arithmetic mean, obtain the broad sense coherence matrix of selected pixel in the master image window;

5b) in auxiliary image window, after being multiplied each other, the broad sense Scattering of Vector of all pixels in this window and the conjugate transpose of self get arithmetic mean, obtain the broad sense coherence matrix of selected pixel in auxiliary image window;

(6) estimate the broad sense interference matrix:

6a) the broad sense Scattering of Vector of all pixels in auxiliary image window is done to conjugate transpose, the broad sense Scattering of Vector of all pixels in the auxiliary image window after the acquisition conjugate transpose;

6b) in the auxiliary image window after conjugate transpose, after the broad sense Scattering of Vector of the broad sense Scattering of Vector of all pixels and the corresponding pixel of master image window is multiplied each other, get arithmetic mean, obtain the broad sense interference matrix;

(7) construction feature split-matrix:

Adopt following formula, the construction feature split-matrix:

Χ=A^-1BC^-1B^H

Wherein, Χ representation feature split-matrix, A means the broad sense coherence matrix of selected pixel in the master image window, B means the broad sense coherence matrix of selected pixel in auxiliary image window, C means the broad sense interference matrix, and subscript-1 representing matrix is inverted, and subscript H means matrix is got to conjugate transpose;

(8) matrix character decomposes:

8a) the feature decomposition matrix is carried out to the feature decomposition operation, obtain the eigenwert of the out of order arrangement of a row feature decomposition matrix;

8b) the feature decomposition matrix is carried out to the feature decomposition operation, obtain row and eigenwert proper vector one to one;

(9) arrayed feature value:

9a) by the eigenwert of the out of order arrangement of feature decomposition matrix, according to being sorted from big to small, obtain the tactic eigenwert of a row feature decomposition matrix;

9b) will with eigenwert proper vector one to one, the order of the eigenwert of arranging is in order sorted, and obtains the tactic proper vector of a row feature decomposition matrix;

(10) generate interferometric phase:

10a) utilize tactic eigenwert and proper vector, the broad sense interference matrix is weighted to processing, generate the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area;

10b) utilize tactic eigenwert and proper vector, the broad sense interference matrix be weighted to processing, generate with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target;

10c) utilize tactic eigenwert and proper vector, the broad sense interference matrix is weighted to processing, generate the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area;

(11) judge whether to obtain all interferometric phases:

Travel through all pixels, judge whether to obtain all interferometric phases corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area, with the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target, with the polarization interference synthetic aperture radar irradiation area in interferometric phase corresponding to tree crown target, if, perform step (12), otherwise, execution step (4);

(12) obtain interferometric phase image:

12a) by the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area of all pixels, be saved in corresponding position in the interferometric phase image corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area, obtain the interferometric phase image corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area;

12b) by all pixels with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target, be saved in in the polarization interference synthetic aperture radar irradiation area between ground and tree crown corresponding position in interferometric phase image corresponding to target, obtain with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase image corresponding to target;

12c) by the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area of all pixels, be saved in corresponding position in the interferometric phase image corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area, obtain the interferometric phase image corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area.

The present invention has the following advantages compared with prior art:

First, the present invention has built the broad sense Scattering of Vector, can be to the adaptive recovery of all polarization informations of current estimation pixel, overcome in prior art based on related weighing associating subspace projection method polarization information utilized to inadequate deficiency, make the present invention have the polarization information utilization fully, interferometric phase image estimates advantage accurately.

Second, the present invention has good robustness to registration error, having overcome prior art utilizes the Cameron decomposition to carry out in the method for polarization interference image registration, the ropy deficiency of interferometric phase image obtained while between POLARIZATION CHANNEL and/or spatial channel, having registration error, make the present invention have advantages of in the situation that POLARIZATION CHANNEL and/or spatial channel exist registration error still can obtain the high-quality interferometric phase image.

The accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention;

Fig. 2 is for generating emulated data topomap used;

Fig. 3 is the interferogram that Pauli base Scattering of Vector method is processed the Generation of simulating data of accuracy registration;

Fig. 4 is the interferogram that the present invention processes the Generation of simulating data of accuracy registration;

Fig. 5 is the interferogram that Pauli base Scattering of Vector method is processed the Generation of simulating data that has 1/10 pixel error;

Fig. 6 is the interferogram that the present invention processes the Generation of simulating data that has 1/10 pixel error;

Fig. 7 is that Pauli base Scattering of Vector method is processed the interferogram that measured data generates;

Fig. 8 is that the present invention processes the interferogram that measured data generates.

Embodiment

Below in conjunction with accompanying drawing, the present invention will be further described.

With reference to accompanying drawing 1, the specific embodiment of the invention step is as follows:

Step 1, input image data.

Main image data and auxiliary view data that the major-minor antenna of polarization interference synthetic aperture radar is obtained respectively are input in system.Main image data and the auxiliary view data of input are the imaging results to areal, with this, guarantee the coherence between master image and auxiliary image.

Step 2, the thick registration of image.

Utilize the thick method for registering of InSAR, master image and auxiliary image are carried out to thick registration process, in the processing procedure of the thick registration of image, registration accuracy is only required and is reached Pixel-level, has greatly alleviated the difficulty of the thick registration of image.

The thick registration concrete steps of InSAR are as follows:

The first step, utilize following formula, calculates the cross-correlation matrix of master image and auxiliary image:

L=IFFT?2(FFT?2(R)×conj(FFT?2(S)))

Wherein, L means the cross-correlation matrix of master image and auxiliary image, and IFFT2 () means to do two-dimentional inverse Fourier transform, FFT2 () means to do two-dimensional Fourier transform, R means main image data, and S means auxiliary view data, and conj () means to get conjugate operation;

Second step, maximizing from all elements of cross-correlation matrix L, obtain coordinate position corresponding to maximal value element in cross-correlation matrix L;

The 3rd step, by the value of the coordinate position of maximal value element in cross-correlation matrix L, deduct the value of the centre coordinate position of cross-correlation matrix L, the side-play amount while obtaining registration;

The 4th step, the equal-sized translation of side-play amount while doing with registration to auxiliary integral image, complete the thick registration of major-minor image.

Step 3, build the broad sense Scattering of Vector.

Building the broad sense Scattering of Vector is in order to utilize all polarization informations of each pixel, around can selecting each pixel and its, 1～8 adjacent pixel builds the broad sense Scattering of Vector, in the embodiment of the present invention, chosen each pixel with its on every side adjacent 8 pixels build the broad sense Scattering of Vector.

The concrete steps that build the broad sense Scattering of Vector are:

3a) choose arbitrarily a pixel in master image, element in the Poly Pauli base Scattering of Vector of 8 pixels adjacent around selected pixel and its is formed a line, construct the broad sense Scattering of Vector of selected pixel in master image;

3b) choose successively the pixel that not yet builds the broad sense Scattering of Vector in master image, by all pixels of choosing successively with the Poly Pauli base Scattering of Vector of adjacent 8 pixels around it in element form a line, construct the broad sense Scattering of Vector of all pixels of choosing in master image;

Whether the broad sense Scattering of Vector that 3c) judges all pixels in master image has all built, if has all built and performed step 3d), otherwise, perform step 3b);

3d) choose arbitrarily a pixel in auxiliary image, element in the Poly Pauli base Scattering of Vector of 8 pixels adjacent around selected pixel and its is formed a line, construct the broad sense Scattering of Vector of selected pixel in auxiliary image;

3e) choose successively the pixel that not yet builds the broad sense Scattering of Vector in auxiliary image, by all pixels of choosing successively with the Poly Pauli base Scattering of Vector of adjacent 8 pixels around it in element form a line, construct the broad sense Scattering of Vector of all pixels of choosing in auxiliary image;

Whether the broad sense Scattering of Vector that 3f) judges all pixels in auxiliary image has all built, if has all built and has performed step 4, otherwise, perform step 3e).

Step 4, choose image window.

In master image, choose a pixel t as center, 3～7 pixels of regular length of take are the piece radius, choose a foursquare window, in auxiliary image, choose with pixel corresponding in master image as center, obtain onesize square window, the square window that to have chosen size in the embodiment of the present invention be 9 * 9.

Step 5, estimate the broad sense coherence matrix.

Under normal conditions, can't directly obtain the assembly average of broad sense coherence matrix, so utilize the space sample of broad sense coherence matrix on average to replace statistical average.

The concrete steps of estimating the broad sense coherence matrix are:

The first step, in the master image window, get arithmetic mean after the broad sense Scattering of Vector of all pixels in this window and the conjugate transpose of self are multiplied each other, and obtains the broad sense coherence matrix Τ of selected pixel in the master image window;

Second step, in auxiliary image window, get arithmetic mean after the broad sense Scattering of Vector of all pixels in this window and the conjugate transpose of self are multiplied each other, and obtains the broad sense coherence matrix Υ of selected pixel in auxiliary image window.

Step 6, estimate the broad sense interference matrix.

Under normal conditions, can't directly obtain the assembly average of broad sense interference matrix, so utilize the space sample of broad sense interference matrix on average to replace statistical average.

The concrete steps of estimating the broad sense interference matrix are:

The first step, do conjugate transpose by the broad sense Scattering of Vector of all pixels in auxiliary image window, the broad sense Scattering of Vector of all pixels in the auxiliary image window after the acquisition conjugate transpose;

Second step, in the auxiliary image window after conjugate transpose, get arithmetic mean after the broad sense Scattering of Vector of the broad sense Scattering of Vector of all pixels and the corresponding pixel of master image window is multiplied each other, and obtains broad sense interference matrix Μ.

Step 7, the construction feature split-matrix.

Adopt following formula, the construction feature split-matrix:

Χ=Τ^-1ΥΜ^-1Υ^H

Wherein, Χ representation feature split-matrix, Τ means the broad sense coherence matrix of the selected pixel of master image window, Υ means the broad sense coherence matrix of the selected pixel of auxiliary image window, Μ means the broad sense interference matrix, and subscript-1 representing matrix is inverted, and H means matrix is got to conjugate transpose.

Step 8, matrix character decomposes.

Feature decomposition matrix X is carried out to the feature decomposition operation, obtain the eigenwert of out of order arrangement of 27 feature decomposition matrixes and 27 and eigenwert proper vector one to one.

Step 9, the arrayed feature value.

The first step, by the eigenwert of the out of order arrangement of feature decomposition matrix, according to being sorted from big to small, obtain the tactic eigenvalue λ of 27 feature decomposition matrixes₁, λ₂..., λ₂₇;

Second step, by 27 with eigenwert proper vector one to one, the order of the eigenwert of arranging is in order sorted, and obtains the tactic proper vector ω of 27 feature decomposition matrixes₁, ω₂..., ω₂₇.

Step 10, generate interferometric phase.

In actual treatment, due to the impact that is subject to the factors such as noise, can't directly obtain accurate eigen vector, can't directly obtain accurate interferometric phase, so the present invention adopts the method for eigenwert and proper vector weighting to estimate interferometric phase.

The concrete steps that generate interferometric phase are as follows:

The first step, adopt following formula, to pixel t, generates the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area:

$\mathrm{\Φ}=\arg (\underset{i=1}{\overset{9}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{\mathrm{\ω}}_i^{H}M{\mathrm{\ω}}_i/\underset{i=1}{\overset{9}{\mathrm{\Σ}}}{\mathrm{\λ}}_i)$

Wherein, Φ means the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area of pixel t, λ_ithe 1st to the 9th eigenwert in the tactic eigenwert of difference representation feature split-matrix, ω_ithe 1st to the 9th proper vector in the tactic proper vector of difference representation feature split-matrix, i=1,2 ... 9, Μ means the broad sense interference matrix, and arg () means to get phase operation, and H means matrix is got to conjugate transpose;

Second step, adopt following formula, to pixel t, generate with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target:

$\mathrm{\Ψ}=\arg (\underset{i=10}{\overset{18}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{\mathrm{\ω}}_i^{H}M{\mathrm{\ω}}_i/\underset{i=10}{\overset{18}{\mathrm{\Σ}}}{\mathrm{\λ}}_i)$

Wherein, Ψ mean pixel t with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target, λ_ithe 10th to the 18th eigenwert in the tactic eigenwert of difference representation feature split-matrix, ω_ithe 10th to the 18th proper vector in the tactic proper vector of difference representation feature split-matrix, i=10,11 ... 18, Μ means the broad sense interference matrix, and arg () means to get phase operation, and subscript H means matrix is got to conjugate transpose;

The 3rd step, adopt following formula, to pixel t, generates the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area:

$\mathrm{\Ω}=\arg (\underset{i=19}{\overset{27}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{\mathrm{\ω}}_i^{H}M{\mathrm{\ω}}_i/\underset{i=19}{\overset{27}{\mathrm{\Σ}}}{\mathrm{\λ}}_i)$

Wherein, Ω means the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area of pixel t, λ_ithe 19th to the 27th eigenwert in the tactic eigenwert of difference representation feature split-matrix, ω_ithe 19th to the 27th proper vector in the tactic proper vector of difference representation feature split-matrix, i=19,20 ... 27, Μ means the broad sense interference matrix, and arg () means to get phase operation, and H means matrix is got to conjugate transpose.

Step 11, judge whether to obtain all interferometric phases.

Travel through all pixels, judge whether to obtain all interferometric phase Φs corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area, with the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase Ψ corresponding to target, with the polarization interference synthetic aperture radar irradiation area in interferometric phase Ω corresponding to tree crown target, if, perform step 12, otherwise, perform step 4.

Step 12, obtain interferometric phase image.

The first step, the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area by all pixels, be saved in corresponding position in the interferometric phase image corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area, obtain the interferometric phase image corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area;

Second step, by all pixels with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target, be saved in in the polarization interference synthetic aperture radar irradiation area between ground and tree crown corresponding position in interferometric phase image corresponding to target, obtain with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase image corresponding to target;

The 3rd step, the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area by all pixels, be saved in corresponding position in the interferometric phase image corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area, obtain the interferometric phase image corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area.

Below in conjunction with emulation experiment, effect of the present invention is described further.

1, simulated conditions:

With reference to accompanying drawing 2, utilize software PolSARPro, to the InSAR simulation imaging that polarized of the scene areas in accompanying drawing 2, obtain respectively the major-minor view data of accuracy registration and have the major-minor view data of 1/10 pixel matching error, simulation parameter arranges as follows:

Parameter	Parameter value
		Theantenna number	2
The main antenna oblique distance	4242.6406m
		The main antenna downwards angle of visibility	45°
Auxiliary antenna oblique distance	4250.4236m
		Auxiliary antenna downwards angle of visibility	45.0858°
Centre frequency	1.3GHz
		Azimuth resolution	1.5m
Slant range resolution	1.0607m
		Average vegetation height	10.0m
Per hectare vegetation density	900
		Orientation is the gradient earthward	0.02
Distance is the gradient earthward	0.01

2, analysis of simulation result:

Fig. 3 (a) means the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area of the Generation of simulating data of Pauli base Scattering of Vector method processing accuracy registration, abscissa axis mean the distance to, unit is pixel, axis of ordinates mean orientation to, unit is pixel, in the image in left side, the gray-scale value of pixel means the size of the interferometric phase of ground scatter corresponding pixel points, the size of interferometric phase fluctuation in figure, reflected the number of phase noise, the image strip on right side means the corresponding relation of gray-scale value in interferometric phase value and left-side images, the size of the numeric representation interferometric phase of image right.Fig. 3 (b) mean Pauli base Scattering of Vector method process accuracy registration Generation of simulating data with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target.Fig. 3 (c) means the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area of the Generation of simulating data of Pauli base Scattering of Vector method processing accuracy registration.Fig. 4 (a) means that the present invention processes the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area of the Generation of simulating data of accuracy registration.Fig. 4 (b) mean the present invention process accuracy registration Generation of simulating data with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target.Fig. 4 (c) means that the present invention processes the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area of the Generation of simulating data of accuracy registration.Fig. 5 (a) means that there be the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area of the Generation of simulating data of 1/10 pixel error in the processing of Pauli base Scattering of Vector method.Fig. 5 (b) mean Pauli base Scattering of Vector method process the Generation of simulating data that has 1/10 pixel error with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target.Fig. 5 (c) means that there be the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area of the Generation of simulating data of 1/10 pixel error in the processing of Pauli base Scattering of Vector method.Fig. 6 (a) means that the present invention processes the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area of the Generation of simulating data that has 1/10 pixel error.Fig. 6 (b) mean the present invention process the Generation of simulating data that has 1/10 pixel error with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target.Fig. 6 (c) means that the present invention processes the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area of the Generation of simulating data that has 1/10 pixel error.

Contrast accompanying drawing 3 and accompanying drawing 4, the fluctuation of the interferometric phase in accompanying drawing 4 will be made a farfetched comparison the little of Fig. 3, and what especially in the zone that has vegetation to cover, manifest is particularly evident, thereby can obtain, when major-minor image accuracy registration, the interferometric phase image quality that the present invention obtains is better than Pauli base Scattering of Vector method.Contrast accompanying drawing 5 and accompanying drawing 6, the fluctuation of the phase value in accompanying drawing 5 is very large, and the phase fluctuation in accompanying drawing 6 is comparatively steady, thereby can obtain, when the registration error of major-minor image is 1/10 pixel, the interferometric phase plot quality that the present invention obtains is apparently higher than Pauli base Scattering of Vector method.Contrast accompanying drawing 4 and accompanying drawing 6, the fluctuation of the interferometric phase value in accompanying drawing 6 is close with the phase fluctuation in accompanying drawing 4, thereby can obtain, and the present invention has good robustness to registration error, in the situation that have registration error between major-minor image, still can obtain quality interferometric phase image preferably.The above analysis, can obtain the present invention when processing polarization InSAR data, and the result obtained is better than existing Pauli base Scattering of Vector method, and registration error is had to good robustness.

3, measured data is processed

While processing measured data, the data of use are the polarization InSAR data that the Amazon zone of Japanese ALOS satellite PALSAR admission obtains.

Fig. 7 (a) means the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area that Pauli base Scattering of Vector method processing measured data generates.Fig. 7 (b) mean that Pauli base Scattering of Vector method processes that measured data generates with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target.Fig. 7 (c) means the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area that Pauli base Scattering of Vector method processing measured data generates.Fig. 8 (a) means that the present invention processes the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area that measured data generates.Fig. 8 (b) mean that the present invention processes that measured data generates with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target.Fig. 8 (c) means that the present invention processes the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area that measured data generates.

Contrast accompanying drawing 7 and accompanying drawing 8, accompanying drawing 8(a) phase fluctuation in is less than accompanying drawing 7(a), accompanying drawing 8(b) phase fluctuation in is significantly less than accompanying drawing 7(b) in phase fluctuation, accompanying drawing 7(c) phase fluctuation in is very large, contain very many phase noises, and accompanying drawing 8(c) in phase fluctuation will to make a farfetched comparison in Fig. 7 (c) phase fluctuation much smaller, can see interference fringe comparatively clearly.As the above analysis, when the present invention processes the polarization InSAR measured data of registration error the unknown, still can obtain than the interferometric phase image of existing disposal route better quality, further verified that the present invention has good robustness to registration error, and practicality of the present invention and validity have been described.

Claims (7)

1. the polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector, comprise the steps:

(1) input image data:

Main image data 1a) the polarization interference synthetic aperture radar main antenna obtained is input to system;

Auxiliary view data 1b) the auxiliary antenna of polarization interference synthetic aperture radar obtained is input to system;

(2) the thick registration of image:

Utilize the thick method for registering of InSAR, master image and auxiliary image are carried out to thick registration process;

(3) build the broad sense Scattering of Vector:

3a) choose arbitrarily a pixel in master image, element in the Poly Pauli base Scattering of Vector of neighbor pixel around selected pixel and its is formed a line, construct the broad sense Scattering of Vector of selected pixel in master image;

3b) choose successively the pixel that not yet builds the broad sense Scattering of Vector in master image, by all pixels of choosing successively with the Poly Pauli base Scattering of Vector of neighbor pixel around it in element form a line, construct the broad sense Scattering of Vector of all pixels of choosing in master image;

Whether the broad sense Scattering of Vector that 3c) judges all pixels in master image has all built, if all built, performs step 3d), otherwise, execution step 3b);

3d) choose arbitrarily a pixel in auxiliary image, element in the Poly Pauli base Scattering of Vector of neighbor pixel around selected pixel and its is formed a line, construct the broad sense Scattering of Vector of selected pixel in auxiliary image;

3e) choose successively the pixel that not yet builds the broad sense Scattering of Vector in auxiliary image, by all pixels of choosing successively with the Poly Pauli base Scattering of Vector of neighbor pixel around it in element form a line, construct the broad sense Scattering of Vector of all pixels of choosing in auxiliary image;

Whether the broad sense Scattering of Vector that 3f) judges all pixels in auxiliary image has all built, if all built, and execution step (4), otherwise, execution step 3e);

(4) choose image window:

In master image, choose a pixel as center, take regular length as the piece radius, choose a foursquare window, in auxiliary image, choose with pixel corresponding in master image as center, obtain onesize square window;

(5) estimate the broad sense coherence matrix:

5a) in the master image window, after being multiplied each other, the broad sense Scattering of Vector of all pixels in this window and the conjugate transpose of self get arithmetic mean, obtain the broad sense coherence matrix of selected pixel in the master image window;

5b) in auxiliary image window, after being multiplied each other, the broad sense Scattering of Vector of all pixels in this window and the conjugate transpose of self get arithmetic mean, obtain the broad sense coherence matrix of selected pixel in auxiliary image window;

(6) estimate the broad sense interference matrix:

6a) the broad sense Scattering of Vector of all pixels in auxiliary image window is done to conjugate transpose, the broad sense Scattering of Vector of all pixels in the auxiliary image window after the acquisition conjugate transpose;

6b) in the auxiliary image window after conjugate transpose, after the broad sense Scattering of Vector of the broad sense Scattering of Vector of all pixels and the corresponding pixel of master image window is multiplied each other, get arithmetic mean, obtain the broad sense interference matrix;

(7) construction feature split-matrix:

Adopt following formula, the construction feature split-matrix:

Χ=A^-1BC^-1B^H

Wherein, Χ representation feature split-matrix, A means the broad sense coherence matrix of selected pixel in the master image window, B means the broad sense coherence matrix of selected pixel in auxiliary image window, C means the broad sense interference matrix, and subscript-1 representing matrix is inverted, and subscript H means matrix is got to conjugate transpose;

(8) matrix character decomposes:

8a) the feature decomposition matrix is carried out to the feature decomposition operation, obtain the eigenwert of the out of order arrangement of a row feature decomposition matrix;

8b) the feature decomposition matrix is carried out to the feature decomposition operation, obtain row and eigenwert proper vector one to one;

(9) arrayed feature value:

9a) by the eigenwert of the out of order arrangement of feature decomposition matrix, according to being sorted from big to small, obtain the tactic eigenwert of a row feature decomposition matrix;

9b) will with eigenwert proper vector one to one, the order of the eigenwert of arranging is in order sorted, and obtains the tactic proper vector of a row feature decomposition matrix;

(10) generate interferometric phase:

10a) utilize tactic eigenwert and proper vector, the broad sense interference matrix is weighted to processing, generate the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area;

10b) utilize tactic eigenwert and proper vector, the broad sense interference matrix be weighted to processing, generate with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target;

10c) utilize tactic eigenwert and proper vector, the broad sense interference matrix is weighted to processing, generate the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area;

(11) judge whether to obtain all interferometric phases:

Travel through all pixels, judge whether to obtain all interferometric phases corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area, with the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target, with the polarization interference synthetic aperture radar irradiation area in interferometric phase corresponding to tree crown target, if, perform step (12), otherwise, execution step (4);

(12) obtain interferometric phase image:

12a) by the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area of all pixels, be saved in corresponding position in the interferometric phase image corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area, obtain the interferometric phase image corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area;

12b) by all pixels with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target, be saved in in the polarization interference synthetic aperture radar irradiation area between ground and tree crown corresponding position in interferometric phase image corresponding to target, obtain with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase image corresponding to target;

12c) by the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area of all pixels, be saved in corresponding position in the interferometric phase image corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area, obtain the interferometric phase image corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area.

2. the polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector according to claim 1, it is characterized in that: the concrete steps of the thick method for registering of InSAR described in step (2) are as follows:

The first step, utilize following formula, calculates the cross-correlation matrix of master image and auxiliary image:

L=IFFT2(FFT?2(R)×conj(FFT?2(S)))

Wherein, L means the cross-correlation matrix of master image and auxiliary image, and IFFT2 () means to do two-dimentional inverse Fourier transform, FFT2 () means to do two-dimensional Fourier transform, R means main image data, and S means auxiliary view data, and conj () means to get conjugate operation;

Second step, maximizing from all elements of cross-correlation matrix L, obtain coordinate position corresponding to maximal value element in cross-correlation matrix L;

The 3rd step, by the value of the coordinate position of maximal value element in cross-correlation matrix L, deduct the value of the centre coordinate position of cross-correlation matrix L, the side-play amount while obtaining registration;

The 4th step, the equal-sized translation of side-play amount while doing with registration to auxiliary integral image, complete the thick registration of major-minor image.

3. the polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector according to claim 1 is characterized in that: the number that is used for building the neighbor pixel of broad sense Scattering of Vector step 3a) is 1～8.

4. the polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector according to claim 1, it is characterized in that: the regular length in step (4) is 3～7 pixels.

5. the polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector according to claim 1 is characterized in that: utilize following formula step 10a), generate the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area:

$\mathrm{\Φ}=\arg (\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{\mathrm{\ω}}_i^{H}M{\mathrm{\ω}}_i/\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\λ}}_i)$

Wherein, Φ means the interferometric phase corresponding with terrain object in the polarization interference synthetic aperture radar irradiation area, the pixel number that m is used while meaning to build the broad sense Scattering of Vector, λ_iin the tactic eigenwert of difference representation feature split-matrix, first is to m eigenwert, ω_irespectively in the tactic proper vector of representation feature split-matrix first to m proper vector, i=1,2 ... m, B means the broad sense interference matrix, and arg () means to get phase operation, and H means matrix is got to conjugate transpose.

6. the polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector according to claim 1, it is characterized in that: utilize following formula step 10b), generate with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target:

$\mathrm{\Ψ}=\arg (\underset{i=m+1}{\overset{2m}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{\mathrm{\ω}}_i^{H}M{\mathrm{\ω}}_i/\underset{i=m+1}{\overset{2m}{\mathrm{\Σ}}}{\mathrm{\λ}}_i)$

Wherein, Ψ mean with in the polarization interference synthetic aperture radar irradiation area between ground and tree crown interferometric phase corresponding to target, the pixel number that m is used while meaning to build the broad sense Scattering of Vector, λ_iin the tactic eigenwert of representation feature split-matrix, m+1 is individual to 2m eigenwert respectively, ω_iin the tactic proper vector of representation feature split-matrix, m+1 is individual to 2m proper vector respectively, i=1, and 2 ... m, B means the broad sense interference matrix, and arg () means to get phase operation, and subscript H means matrix is got to conjugate transpose.

7. the polarization InSAR interferogram method of estimation based on the broad sense Scattering of Vector according to claim 1 is characterized in that: utilize following formula step 10c), generate the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area:

$\mathrm{\Ω}=\arg (\underset{i=m+1}{\overset{2m}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{\mathrm{\ω}}_i^{H}M{\mathrm{\ω}}_i/\underset{i=m+1}{\overset{2m}{\mathrm{\Σ}}}{\mathrm{\λ}}_i)$

Wherein, Ω means the interferometric phase corresponding with tree crown target in the polarization interference synthetic aperture radar irradiation area, the pixel number that m is used while meaning to build the broad sense Scattering of Vector, λ_iin the tactic eigenwert of representation feature split-matrix, 2m+1 is individual to 3m eigenwert respectively, ω_iin the tactic proper vector of representation feature split-matrix, 2m+1 is individual to 3m proper vector respectively, i=1, and 2 ... m, B means the broad sense interference matrix, and arg () means to get phase operation, and H means matrix is got to conjugate transpose.

Images